Easy Distributed Systems Using Hazelcast. Peter Veentjer

Transcription

1 Easy Distributed Systems Using Hazelcast Peter Veentjer

2 Whoami Working for Hazelcast Team Lead Quality & Performance Author of Mastering Hazelcast 3' 14 years Java experience Big love Concurrency control Distributed computing

3

4 What is Hazelcast?

5 What is Hazelcast? Leading Open Source Java In Memory Data/ Compute Grid Out main goal is to simplify development of: Scalable systems Highly available systems

6 Why Hazelcast? 3 MByte JAR no other dependencies! no need to install software! Its a library, not an application framework Apache 2 License Free to use no limitations

7 Distributed Data-structures IAtomicLong/Ref IdGenerator Lock/Condition CountDownLatch Semaphore Queue Map MultiMap Set List Topic Executor ReplicatedMap MapReduce TransactionalMap/Q/S Write your own!

8 So what can I do with it?

9 Scaling

10 Scale up

11 Scale Out

12 High Availability

13 read write

14 Raspberry Pi Cluster

15 When? Caching Messaging Solution Data Aggregation Clustered Scheduling Job Processing Cluster management HTTP session clustering

16 Which companies E-Commerce Apple, ebay Financials JP Morgan, Morgan Stanley, HSBC, Deutsche Bank Telco s AT&T, Ericsson Gaming Ubisoft/Blue Byte

17 Which Open Source Projects WSO2 Carbon Mule ESB Vert.x Apache Camel Apache Shiro OrientDB Alfresco Karaf

18 How Often In April M startups 48K unique

19 Where is the code!

20 Creating Hazelcast Cluster HazelcastInstance hz = Hazelcast.newHazelcastInstance();

21 XML Configuration <hazelcast> <network> </network> <map name= m > </map> <queue name= q > </queue> < </hazelcast>

22 Programmatic Configuration Config config = new Config(); make config modifications HazelcastInstance hz = Hazelcast.newHazelcastInstance(config);

23 Demo

24 Map Map<String,String> products = new HashMap(); map.put("1","iphone");

25 Map Map<String,String> products = new ConcurrentHashMap(); map.put("1","iphone");

26 Map HazelcastInstance hz = Hazelcast.newHazelcastInstance(); Map<String,String> products = hz.getmap("products"); cities.put("1","iphone");

27 Demo

28 Map Configuration! <map name="products"> <time-to-live-seconds>4</time-to-live-seconds> <indexes> <index>name</index> </indexes> <..> </map>

29 Map Scalability and High Availability 271 Partitions partitionid = hash(key) % partitioncount

30 Map Scalability and High Availability Member 1

31 Map Scalability and High Availability Member 1 Member 2

32 Map Scalability and High Availability Member 1 Member 2 Member 3

33 Map Scalability and High Availability Member 1 Member 2 Member 3

34 Map Scalability and High Availability Member 1 Member 2 Member 3

35 Map Scalability and High Availability Member 1 Member 2

36 Demo

37 Multimap Collection as value Serialization Overhead Lost update

38 Demo

39 Queue BlockingQueue queue = new LinkedBlockingQueue(); queue.offer("1"); Object item = queue.take();

40 Queue! HazelcastInstance hz = Hazelcast.newHazelcastInstance(); BlockingQueue queue = hz.getqueue("queue"); queue.offer("1"); Object item = queue.take();

41 Demo

42 Topic ITopic topic = hz.gettopic("topic"); //publishing topic.publish(msg);! //subscribing topic.addmessagelistener(new TopicSubscriber());! public class TopicSubscriber implements MessageListener<String> { public void onmessage(message<string> m) { System.out.println(m.getMessageObject()); } }

43 Client Languages Java C++ C# Memcached Smart vs Dumb

44 Client HazelcastInstance client = HazelcastClient.newHazelcastClient();

45 Network Communication Cluster Discovery Multicast TCP/IP Cluster AWS Cluster Normal Network Communication TCP/IP

46 Advanced Hazelcast

47 Data Locality: Problem! ILock lock = hz.getlock( somelock );! IAtomicLong counter =hz.getatomiclong( somecounter );! IMap map = hz.getmap( somemap ); map.put( someperson,new Person()); map.get( someperson );

48 Member 1 SomeLock Member 2 SomeCounter Member 3 SomePerson

49 name

50 Data Locality: Fixed ILock lock = hz.getlock( somelock@foo ); IAtomicLong counter = hz.getatomiclong( somecounter@foo );! IMap map = hz.getmap( somemap ); map.put( someperson@foo, new Person()); Person p = map.get( someperson@foo );

51 Member 1 SomeLock Foo Partition Member 2 Member 3 SomeCounter SomePerson

52 Adding object in same partition IAtomicLong c1 = IAtomicLong c2 = hz.getatomiclong( c2@ +c1.getpartitionkey());

53 Executor Execute task anywhere Execute task on key owner Execute task one/all/subset members Synchronisation Future ExecutionCallback CompletableFuture Hazelcast 3.3

54 Demo

55 Locking Locks TransactionalMap.getForUpdate Map.Lock

56 Demo

57 Race problem public void increment(int accountid,int amount){ Account account = accounts.get(accountid); account.balance+=amount; accounts.put(accountid, account); }

58 Pessimistic Increment public void increment(int accountid,int amount){ accounts.lock(accountid); try{ Account account = accounts.get(accountid); account.balance+=amount; accounts.put(accountid, account); }finally{ accounts.unlock(accountid); } }

59 Pessimistic Increment public void increment(int accountid,int amount){ accounts.lock(accountid); try{ Account account = accounts.get(accountid); account.balance+=amount; accounts.put(accountid, account); }finally{ accounts.unlock(accountid); } }

60 Optimistic Increment public void increment(int accountid,int amount){ for(;;){ Account old = accounts.get(accountid); Account update = new Account(old); update.balance+=amount; if(accounts.replace(accountid, old,update)){ return; } } }

61 Optimistic Increment public void increment(int accountid,int amount){ for(;;){ Account old = accounts.get(accountid); Account update = new Account(old); update.balance+=amount; if(accounts.replace(accountid, old,update)){ return; } } }

62 Read Data Write Data DATA Bad: Send Data to Function Write Function DATA Good: Send Function to Data

63 class BalanceTask implements Runnable,Serializable{ private int accountid, amount; public void run() { for(;;){ Account old = accounts.get(accountid); Account update = new Account(old); update.balance+=amount; if(accounts.replace(accountid, old,update)){ return; } } } }

64 Increment with runnable public void increment(int id, int amount){ BalanceTask task = new BalanceTask(id,amount); executorservice.executeonkeyowner(task,id); }

65 Increment with EntryProcessor class BalanceProcessor extends AbstractEntryProcessor{ int amount; BalanceProcessor(int amount) { this.amount = amount; }! Object process(map.entry<integer, Account> e) { e.getvalue().balance+=amount; return null; }

66 Using entry processor public void increment(int accountid, int amount){ BalanceProcessor p = new BalanceProcessor(amount); accounts.executeonkey(accountid, p); }

67 Map: In Memory Format 2 Options BINARY OBJECT Predicates EntryProcessor

68 SPI You can write your own distributed datastructures Example Distributed Actors implementation

69 Serialization API Serializable/Externalizable Portable ByteArraySerializable StreamSerializer Kryo Jackson Smile Protobuf

70 Kryo Serialization Serialization 20,000 Size in Bytes 15,000 10,000 5,000 0 Kryo Kryp + Compr

71 Pluggable Serialization! SerializerConfig ps = new SerializerConfig().setTypeClass(Product.class).setImplementation(new ProductSerializer());! Config config = new Config(); config.getserializationconfig().addserializerconfig(ps);! HazelcastInstance hz = Hazelcast.newHazelcastInstance(config);

72 Hazelcast 3.2 Map Reduce IMap<Integer,Account> accounts = hz.getmap( accounts );! Map<Integer,Integer> result = accounts.map(new AccountSumMapper()).reduce(new AccountSumReducer()).submit()

73 Enterprise vs Community edition Support contracts Elastic Memory Security Management Center

74 The Future Stabilizer Topologies Dynamic Creation Map of Maps Tiered storage

75 Questions?